CN116543914A - Model construction method for evaluating second molar tooth root absorption degree based on CBCT image - Google Patents
Model construction method for evaluating second molar tooth root absorption degree based on CBCT image Download PDFInfo
- Publication number
- CN116543914A CN116543914A CN202310531653.1A CN202310531653A CN116543914A CN 116543914 A CN116543914 A CN 116543914A CN 202310531653 A CN202310531653 A CN 202310531653A CN 116543914 A CN116543914 A CN 116543914A
- Authority
- CN
- China
- Prior art keywords
- molar
- resnet50
- cbam
- absorption
- grading
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010521 absorption reaction Methods 0.000 title claims abstract description 69
- 238000007408 cone-beam computed tomography Methods 0.000 title claims abstract description 44
- 238000010276 construction Methods 0.000 title claims abstract description 9
- 210000004357 third molar Anatomy 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 11
- 238000011156 evaluation Methods 0.000 claims abstract description 10
- 238000003066 decision tree Methods 0.000 claims abstract description 5
- 238000013480 data collection Methods 0.000 claims abstract description 4
- 238000000034 method Methods 0.000 claims description 16
- 210000004513 dentition Anatomy 0.000 claims description 13
- 230000036346 tooth eruption Effects 0.000 claims description 13
- 238000011176 pooling Methods 0.000 claims description 12
- 230000003213 activating effect Effects 0.000 claims description 3
- 210000003074 dental pulp Anatomy 0.000 claims description 3
- 210000004262 dental pulp cavity Anatomy 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 208000006860 root resorption Diseases 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims 1
- 230000000007 visual effect Effects 0.000 claims 1
- 230000006870 function Effects 0.000 description 6
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 4
- 210000000988 bone and bone Anatomy 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 239000007943 implant Substances 0.000 description 2
- 238000012800 visualization Methods 0.000 description 2
- HRANPRDGABOKNQ-ORGXEYTDSA-N (1r,3r,3as,3br,7ar,8as,8bs,8cs,10as)-1-acetyl-5-chloro-3-hydroxy-8b,10a-dimethyl-7-oxo-1,2,3,3a,3b,7,7a,8,8a,8b,8c,9,10,10a-tetradecahydrocyclopenta[a]cyclopropa[g]phenanthren-1-yl acetate Chemical compound C1=C(Cl)C2=CC(=O)[C@@H]3C[C@@H]3[C@]2(C)[C@@H]2[C@@H]1[C@@H]1[C@H](O)C[C@@](C(C)=O)(OC(=O)C)[C@@]1(C)CC2 HRANPRDGABOKNQ-ORGXEYTDSA-N 0.000 description 1
- 208000003556 Dry Eye Syndromes Diseases 0.000 description 1
- 206010013774 Dry eye Diseases 0.000 description 1
- 208000017442 Retinal disease Diseases 0.000 description 1
- 206010038923 Retinopathy Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000009933 burial Methods 0.000 description 1
- 230000001055 chewing effect Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 206010012601 diabetes mellitus Diseases 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 210000000214 mouth Anatomy 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 238000004393 prognosis Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000001356 surgical procedure Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 210000001519 tissue Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention provides a model construction method for evaluating the absorption degree of a tooth root of a second molar based on a CBCT image, which comprises the following processing steps of (1) CBCT image data collection and pretreatment; (2) res net50 network training; (3) ResNet50-CBAM network training: introducing a convolution block attention module CBAM into a ResNet50 grading model to form a ResNet50-CBAM network, inputting the information of grading the second molar tooth root absorption cross-section image obtained in the step (1) according to the absorption degree into the ResNet50-CBAM network, training according to a set grading mode to obtain a ResNet50-CBAM grading model for evaluating the external absorption of the second molar tooth root, and inputting grading results obtained by the ResNet50-CBAM grading model into a decision tree so as to output an evaluation result of whether to extract the third molar tooth. The invention can be effectively used for evaluating the external absorption of the tooth root of the second molar, and provides reference comments for clinicians.
Description
Technical Field
The invention belongs to the field of biological information, and particularly relates to a model construction method for evaluating the absorption degree of a second molar tooth root based on a CBCT image.
Background
Third Molar (M3) is usually the latest erupting tooth in human dentition, and when human grows to age 17-21 years, it erupts at the distal middle end of Second Molar (M2), but because of gradual diversification and refinement of human food species, deterioration of masticatory organ function, the space at the erupting site of M3 is insufficient, and in addition, hard tissues are all around the Third Molar, which are easily blocked and cannot erupt normally, and thus, the blocking occurs. Bone-embedded third molar (Bone-impacted third molar, BI-M3) is one form of low steric hindrance, as it is fully embedded in Bone, with a relatively low relative position, often in contact with the root of the second molar of the adjacent tooth. Mediating secondary molar root resorption may even cause surrounding bone destruction, affecting tooth chewing function, dental stability and even dentition stability, requiring assessment of secondary molar root resorption.
With the development of big data and computer technology, artificial intelligence is an effective solution. Convolutional Neural Networks (CNNs) are the core of artificial intelligence, consisting of neurons self-optimized by learning, composed of outputs, with multiple hidden layers in between. The main applications in the medical field are semantic segmentation, feature extraction and hierarchical diagnosis. With the development of deep learning, it also has excellent performance in disease evaluation. If the characteristics extracted by the convolutional neural network are used as input, the support vector machine is used for grading, so that the diagnosis of the retinopathy image caused by diabetes can be realized. Also, the learner uses a deep convolutional neural network model to detect fluorescent tear film rupture zones, by sliding window detection to screen the rupture center and to derive tear film rupture time. This method has been used to evaluate the stable progression of the tear film and automatically evaluate the condition of dry eye. In the aspect of oral cavity, as in CN115512167A, a method and a device for constructing a classification model for evaluating the stability of an implant based on CBCT image data are used for evaluating the stability of the implant. While intelligent natural tooth assessment schemes for molar teeth are currently unavailable. At present, the problem of 'prognosis of M2 extradental absorption under the influence of BI-M3' is not clear, and a systematic evaluation method is also lacked. In addition, the clinician can only rely on subjective experience and trade-off in making a decision of whether to pull out BI-M3, and guidance under objective data support is not available. Thus, a model building method for evaluating the second molar root absorption based on CBCT images is needed to meet the needs of use.
Disclosure of Invention
The invention aims to provide a model construction method for evaluating the absorption degree of a second molar tooth root based on CBCT images, which can be effectively used for evaluating the external absorption of the second molar tooth root and provides reference comments for clinicians.
To achieve the object, there is provided a model construction method for evaluating the degree of root absorption of a second molar tooth based on CBCT images, the method comprising the steps of;
(1) CBCT image data collection and preprocessing: collecting CBCT data of a plurality of patients, processing the collected CBCT data, reconstructing to obtain a dentition cross-section image, and marking the absorption degree of the second molar M2 in the cross-section image according to different grades according to preset standards;
(2) ResNet50 network training: taking the ResNet50 network as a trunk model, inputting the information of grading the absorption cross section image of the second molar tooth root obtained in the step (1) according to the absorption degree into the ResNet50 network, and training according to a preset grading mode to obtain a ResNet50 grading model for evaluating the external absorption of the second molar tooth root;
(3) ResNet50-CBAM network training: introducing a convolution block attention module CBAM into a ResNet50 grading model to form a ResNet50-CBAM network, inputting the information of grading the second molar tooth root absorption cross-section image obtained in the step (1) according to the absorption degree into the ResNet50-CBAM network, training according to a set grading mode to obtain a ResNet50-CBAM grading model for evaluating the external absorption of the second molar tooth root, and inputting grading results obtained by the ResNet50-CBAM grading model into a decision tree so as to output an evaluation result of whether to extract the third molar tooth.
Preferably, in step (1), the CBCT image data is scanned by NewTomVG and the resulting image data is stored as DICOM data.
Preferably, in step (1), the CBCT image data is reconstructed in NewTomVG software according to a dentition curve, cross-sectional images perpendicular to the curve are reconstructed at a thickness of 0.2-0.4mm, dentition cross-sectional images are obtained and read grading and marking are performed by a physician according to preset criteria.
Preferably, in step (1), the classification of the preset criteria is: the second molar is classified into five grades of Normal, level 1, level 2, level 3 and Level 4 according to the relation between the third molar and the second molar and the different absorption degrees of the tooth roots of the second molar, wherein the Normal is as follows: the third molar is fully bone-embedded, but is not in contact with the second molar; level 1: the third molar is in contact with the second molar, but the second molar root is not resorbed or the cementum is slightly incomplete; level 2: the third molar is contacted with the second molar, the second molar is slightly absorbed, and the external absorption of the tooth root is less than half of the thickness of the root canal wall; level 3: the third molar is contacted with the second molar, which is moderately absorbed, and the extradental absorption is greater than half the thickness of the canal wall but does not accumulate dental pulp; level 4: the third molar is in contact with the second molar, which is heavily absorbed, and extraroot absorbed accumulated pulp.
Preferably, in step (2), the ResNet50 network is trained through a large visualization database ImageNet.
Preferably, in step (3), the CBCT image in step (1) is batch-converted into a jpg format file by using a medical image processing tool pydicom in the python procedure, an image is taken with a thickness of 0.3mm as a plane, one CBCT image horizontal plane and sagittal plane are displayed by using an opencv library, a doctor selects a clipping range to determine a rectangular upper left corner coordinate (x 1, y 1) and a rectangular lower right corner coordinate (x 2, y 2) to obtain a region of interest in the image, and then the size of each image is adjusted to 224 x 224 pixels; and then acquiring the whole image sequence intercepted by the patient based on the coordinates, and storing the sequence summary set as an independent jpg file to obtain a second molar tooth root absorption cross-section image, inputting information of grading the second molar tooth root absorption cross-section image according to the absorption degree into a ResNet50-CBAM network, and training according to the set grading information to obtain a ResNet50-CBAM grading model for evaluating the external absorption of the second molar tooth root.
Preferably, in step (3), a convolutional block attention module CBAM is introduced into the res net50 hierarchical model to form a res net50-CBAM network, wherein the convolutional block attention module CBAM is disposed in a layer of the res net50 hierarchical model.
Preferably, the convolution block attention module CBAM is provided with a channel attention module CAM and a spatial attention module SAM, where the channel attention module CAM is configured to perform maximum pooling and average pooling processing on input features in a channel dimension, then input pooled data to a multi-layer perceptron MLP with shared parameters, and connect with each other and be activated by an S-type function sigmoid in a nonlinear manner, that is, the input features of the channel attention module CAM are multiplied by pooled output features to obtain channel refinement features; the space attention module SAM is used for carrying out maximum pooling and average pooling treatment on the channel refinement characteristics obtained by the channel attention module CAM in the space dimension, and activating by a 7 multiplied by 7 convolution kernel and an S-type function sigmoid to obtain SAM output; the CBAM refinement feature is obtained by multiplying the channel refinement feature with the output of the SAM.
Preferably, in the ResNet50-CBAM network, the ResNet50 comprises four layers, namely, 3, 4, 6 and 3 bottleneck layers in layer1, layer2, layer3 and layer4, respectively, in each layer, each of the bottleneck layers is arranged in sequence, and one CBAM is arranged among the layers, so that the CBAM module is added to the ResNet 50.
Compared with the prior art, the invention has the beneficial effects that:
in the invention, the ResNet50-CBAM network is formed by introducing the ResNet50 hierarchical model into the convolution block attention module CBAM to evaluate the cross-sectional image of the second molar root absorption, so that the method can be effectively used for evaluating the second molar root external absorption, and provides reference comments for clinicians. According to the invention, the evaluation result of whether to pull out the third molar is obtained, so that the clinical doctors can be helped to make more accurate diagnosis, the reading time of the clinical doctors is reduced, the pressure of the doctors in work is relieved, and in addition, the device can be used as a teaching aid to train the diagnostic ability of the oral maxillofacial surgery students.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a flow chart of the ResNet50-CBAM hierarchical model operation in the present invention;
fig. 3 is a schematic view showing the second molar absorption degree classification according to the present invention.
Detailed Description
The technical solution of the present invention will be described in further detail with reference to the accompanying drawings, but the scope of the present invention is not limited to the following description.
1-3, the present invention provides a model construction method for evaluating the tooth root absorption degree of a second molar based on CBCT images, comprising the following processing steps;
(1) CBCT image data collection and preprocessing: collecting CBCT data of a plurality of patients, processing the collected CBCT data, reconstructing to obtain a dentition cross-section image, and marking the absorption degree of the second molar M2 in the cross-section image according to different grades according to preset standards;
(2) ResNet50 network training: taking the ResNet50 network as a trunk model, inputting the information of grading the absorption cross section image of the second molar tooth root obtained in the step (1) according to the absorption degree into the ResNet50 network, and training according to a preset grading mode to obtain a ResNet50 grading model for evaluating the external absorption of the second molar tooth root;
(3) ResNet50-CBAM network training: introducing a convolution block attention module CBAM into a ResNet50 grading model to form a ResNet50-CBAM network, inputting the information of grading the second molar tooth root absorption cross-section image obtained in the step (1) according to the absorption degree into the ResNet50-CBAM network, training according to a set grading mode to obtain a ResNet50-CBAM grading model for evaluating the external absorption of the second molar tooth root, and inputting grading results obtained by the ResNet50-CBAM grading model into a decision tree so as to output an evaluation result of whether to extract the third molar tooth.
In step (1), CBCT image data is scanned by NewTomVG and the resulting image data is stored as DICOM data.
In step (1), the CBCT image data is reconstructed in NewTomVG software according to a dentition curve, cross-sectional images perpendicular to the curve are reconstructed at a thickness of 0.2-0.4mm, dentition cross-sectional images are obtained and read grading and marking are performed by a physician according to preset criteria. The CBCT image data collected patients met the requirements of being older than eighteen years and less than sixty years, third molar full bone burial, and, being healthy throughout the body and free of underlying disease. All patient personally identifiable information is first deleted by an anonymous process. The NewTomVG parameter settings at the time of CBCT image data acquisition are as shown in table 1.
Table 1: CBCT image shooting parameters
In step (1), the classification of the preset criteria is: the second molar is classified into five grades of Normal, level 1, level 2, level 3 and Level 4 according to the relation between the third molar and the second molar and the different absorption degrees of the tooth roots of the second molar, wherein the Normal is as follows: the third molar is fully bone-embedded, but is not in contact with the second molar; level 1: the third molar is in contact with the second molar, but the second molar root is not resorbed or the cementum is slightly incomplete; level 2: the third molar is contacted with the second molar, the second molar is slightly absorbed, and the external absorption of the tooth root is less than half of the thickness of the root canal wall; level 3: the third molar is contacted with the second molar, which is moderately absorbed, and the extradental absorption is greater than half the thickness of the canal wall but does not accumulate dental pulp; level 4: the third molar is in contact with the second molar, which is heavily absorbed, and extraroot absorbed accumulated pulp. The M2-ERR levels were varied according to dentition cross-section, level 1A/B, level 2C/D, level 3E/F, level 4G/H. In this embodiment, the Decision tree Decision-making tree is internally provided with a parameter for outputting whether to pick up the third molar according to the external root absorption evaluation result of the second molar of the patient, for example, when Level 4 is used for picking up the third molar.
In step (2), the ResNet50 network is trained over the large visualization database ImageNet.
In the step (3), the CBCT image in the step (1) is converted into a jpg format file in batches by using a medical image processing tool pydicom in a python program, an image is intercepted by taking 0.3mm as a plane, then the plane and the sagittal plane of one CBCT image are displayed by using an opencv library, a doctor selects a clipping range to determine the upper left corner coordinates (x 1, y 1) and the lower right corner coordinates (x 2, y 2) of a rectangle so as to acquire an interest area in the image, and then the size of each image is adjusted to 224 x 224 pixels; and then acquiring the whole image sequence intercepted by the patient based on the coordinates, and storing the sequence summary set as an independent jpg file to obtain a second molar tooth root absorption cross-section image, inputting information of grading the second molar tooth root absorption cross-section image according to the absorption degree into a ResNet50-CBAM network, and training according to the set grading information to obtain a ResNet50-CBAM grading model for evaluating the external absorption of the second molar tooth root.
In step (3), a convolution block attention module CBAM is introduced into the ResNet50 hierarchical model to form a ResNet50-CBAM network, wherein the convolution block attention module CBAM is arranged in a layer of the ResNet50 hierarchical model.
The convolution block attention module CBAM is provided with a channel attention module CAM and a space attention module SAM, wherein the channel attention module CAM is used for carrying out maximum pooling and average pooling processing on input features in a channel dimension, then pooled data is input into a multi-layer perceptron MLP with shared parameters, and the pooled data are connected with each other and are activated by sigmoid nonlinearity of an S-shaped function, namely, the input features of the channel attention module CAM are multiplied by pooled output features to obtain channel refinement features; the space attention module SAM is used for carrying out maximum pooling and average pooling treatment on the channel refinement characteristics obtained by the channel attention module CAM in the space dimension, and activating by a 7 multiplied by 7 convolution kernel and an S-type function sigmoid to obtain SAM output; the CBAM refinement feature is obtained by multiplying the channel refinement feature with the output of the SAM.
In the ResNet50-CBAM network, the ResNet50 comprises four layers, namely, 3, 4, 6 and 3 bottleneck layers in layer1, layer2, layer3 and layer4, respectively, in each layer, each bottleneck is arranged in sequence, and one CBAM is arranged among the bottlenecks, so that the CBAM module is added to the ResNet 50.
In this embodiment, through 360 cases of CBCT images of the affected teeth, 5275 dentition cross-sectional images are obtained, 1648 images which meet the classification standard are taken into the dataset, and are divided into a training set and a test set according to the ratio of 8:2. The accuracy of the ResNet50-CBAM model is 0.9407, and the ResNet50-CBAM model is 0.8204 in terms of sensitivity; in the aspect of specificity, the ResNet50-CBAM model is 0.9553, so that the invention has higher diagnosis performance, short time consumption of evaluation and high accuracy, can be effectively used for the evaluation of the external absorption of the tooth root of the second molar, and provides reference comments for clinicians.
The foregoing is merely a preferred embodiment of the invention, and it is to be understood that the invention is not limited to the form disclosed herein and is not to be considered as excluding other embodiments, but is capable of numerous other combinations, modifications and environments and is capable of modifications within the scope of the inventive concept, either by the foregoing teachings or by the teaching of the relevant art. And that modifications and variations which do not depart from the spirit and scope of the invention are intended to be within the scope of the appended claims.
Claims (9)
1. A model construction method for evaluating the absorption degree of a tooth root of a second molar based on a CBCT image, which is characterized by comprising the following processing steps of;
(1) CBCT image data collection and preprocessing: collecting CBCT data of a plurality of patients, processing the collected CBCT data, reconstructing to obtain a dentition cross-section image, and marking the absorption degree of the second molar M2 in the cross-section image according to different grades according to preset standards;
(2) ResNet50 network training: taking the ResNet50 network as a trunk model, inputting the information of grading the absorption cross section image of the second molar tooth root obtained in the step (1) according to the absorption degree into the ResNet50 network, and training according to a preset grading mode to obtain a ResNet50 grading model for evaluating the external absorption of the second molar tooth root;
(3) ResNet50-CBAM network training: introducing a convolution block attention module CBAM into a ResNet50 grading model to form a ResNet50-CBAM network, inputting the information of grading the second molar tooth root absorption cross-section image obtained in the step (1) according to the absorption degree into the ResNet50-CBAM network, training according to a set grading mode to obtain a ResNet50-CBAM grading model for evaluating the external absorption of the second molar tooth root, and inputting grading results obtained by the ResNet50-CBAM grading model into a decision tree so as to output an evaluation result of whether to extract the third molar tooth.
2. A method of constructing a model for assessing the extent of root absorption of a second molar tooth based on CBCT images according to claim 1 wherein in step (1) the CBCT image data is scanned by NewTomVG and the resulting image data is stored as DICOM data.
3. A method of constructing a model for assessing the extent of tooth root absorption of a second molar based on CBCT images according to claim 1 or 2, wherein in step (1) the CBCT image data is reconstructed in NewTomVG software according to a dentition curve, cross-sectional images perpendicular to the curve are reconstructed at a thickness of 0.2-0.4mm, dentition cross-sectional images are obtained and read grading and labeling by a physician according to preset criteria.
4. A method of modeling for assessing a degree of root absorption of a second molar tooth based on CBCT images according to claim 3, wherein in step (1), the classification of the predetermined criteria is: the second molar is classified into five grades of Normal, level 1, level 2, level 3 and Level 4 according to the relation between the third molar and the second molar and the different absorption degrees of the tooth roots of the second molar, wherein the Normal is as follows: the third molar is fully bone-embedded, but is not in contact with the second molar; level 1: the third molar is in contact with the second molar, but the second molar root is not resorbed or the cementum is slightly incomplete; level 2: the third molar is contacted with the second molar, the second molar is slightly absorbed, and the external absorption of the tooth root is less than half of the thickness of the root canal wall; level 3: the third molar is contacted with the second molar, which is moderately absorbed, and the extradental absorption is greater than half the thickness of the canal wall but does not accumulate dental pulp; level 4: the third molar is in contact with the second molar, which is heavily absorbed, and extraroot absorbed accumulated pulp.
5. A method of modeling for assessing a degree of root resorption of a second molar based on a CBCT image according to claim 1, wherein in step (2) the res net50 network is trained through a large visual database ImageNet.
6. A method of constructing a model for assessing the extent of tooth root absorption of a second molar based on CBCT images according to claim 1 or 3, wherein in step (3), the CBCT images in step (1) are batch-converted into a jpg format file using the medical image processing tool pydicom in the python procedure, the images are taken with a thickness of 0.3mm as a plane, one of the CBCT image horizontal plane and sagittal plane is displayed using the opencv library and a clipping range is selected by a doctor to determine the upper left angular coordinates (x 1, y 1) and the lower right angular coordinates (x 2, y 2) of the rectangle to obtain a region of interest in the image, and then the size of each image is adjusted to 224 x 224 pixels; and then acquiring the whole image sequence intercepted by the patient based on the coordinates, and storing the sequence summary set as an independent jpg file to obtain a second molar tooth root absorption cross-section image, inputting information of grading the second molar tooth root absorption cross-section image according to the absorption degree into a ResNet50-CBAM network, and training according to the set grading information to obtain a ResNet50-CBAM grading model for evaluating the external absorption of the second molar tooth root.
7. The method of constructing a model for assessing a degree of root absorption of a second molar tooth based on a CBCT image according to claim 6, wherein in step (3), a convolutional block attention module CBAM is introduced into the ResNet50 hierarchical model to form a ResNet50-CBAM network, wherein the convolutional block attention module CBAM is disposed in a layer of the ResNet50 hierarchical model.
8. The method for constructing a model for evaluating the absorption degree of the tooth root of the second molar based on the CBCT image according to claim 7, wherein the convolution block attention module CBAM is provided with a channel attention module CAM and a spatial attention module SAM, wherein the channel attention module CAM is configured to perform a maximum pooling and an average pooling process on the input feature in the channel dimension, then input the pooled data to a multi-layer perceptron MLP with shared parameters, and then connect with each other and be activated by sigmoid nonlinearity, i.e. the input feature of the channel attention module CAM is multiplied by the pooled output feature to obtain a channel refinement feature; the space attention module SAM is used for carrying out maximum pooling and average pooling treatment on the channel refinement characteristics obtained by the channel attention module CAM in the space dimension, and activating by a 7 multiplied by 7 convolution kernel and an S-type function sigmoid to obtain SAM output; the CBAM refinement feature is obtained by multiplying the channel refinement feature with the output of the SAM.
9. The model construction method for evaluating the tooth root absorption degree of the second molar tooth based on the CBCT image according to claim 7, wherein in the ResNet50-CBAM network, the ResNet50 comprises four layers, namely, 3, 4, 6 and 3 bottleneck layers in layer1, layer2, layer3 and layer4 respectively, in each layer, each of the bottlenecks is arranged in sequence, and one CBAM is arranged among the bottlenecks, so that the CBAM module is added to the ResNet 50.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310531653.1A CN116543914A (en) | 2023-05-12 | 2023-05-12 | Model construction method for evaluating second molar tooth root absorption degree based on CBCT image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310531653.1A CN116543914A (en) | 2023-05-12 | 2023-05-12 | Model construction method for evaluating second molar tooth root absorption degree based on CBCT image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116543914A true CN116543914A (en) | 2023-08-04 |
Family
ID=87450178
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310531653.1A Pending CN116543914A (en) | 2023-05-12 | 2023-05-12 | Model construction method for evaluating second molar tooth root absorption degree based on CBCT image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116543914A (en) |
-
2023
- 2023-05-12 CN CN202310531653.1A patent/CN116543914A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Agrawal et al. | Artificial intelligence in dentistry: past, present, and future | |
| Ng et al. | Computerized detection of breast cancer with artificial intelligence and thermograms | |
| US20160038092A1 (en) | Applying non-real time and non-user attended algorithms to stored non-imaging data and existing imaging data for obtaining a dental diagnosis | |
| CN107066814A (en) | A kind of traditional Chinese medical science intelligent auxiliary diagnosis system cooperateed with based on the four methods of diagnosis | |
| CN112037913B (en) | Periodontitis intelligent detection system based on convolutional neural network | |
| CN110503652A (en) | Mandibular kinesiography and adjacent teeth and mandibular canal relationship determine method, apparatus, storage medium and terminal | |
| Babu et al. | Artificial intelligence in dentistry: concepts, applications and research challenges | |
| Nie et al. | OII-DS: A benchmark Oral Implant Image Dataset for object detection and image classification evaluation | |
| Braga et al. | The enamel–dentine junction in the postcanine dentition of Australopithecus africanus: intra‐individual metameric and antimeric variation | |
| Milošević et al. | A comprehensive exploration of neural networks for forensic analysis of adult single tooth X-Ray images | |
| Manic et al. | Jaya algorithm-assisted evaluation of tooth elements using digital bitewing radiography images | |
| Tian et al. | Efficient tooth gingival margin line reconstruction via adversarial learning | |
| Datta et al. | Neutrosophic set-based caries lesion detection method to avoid perception error | |
| CN116543914A (en) | Model construction method for evaluating second molar tooth root absorption degree based on CBCT image | |
| CN114004970A (en) | Tooth area detection method, device, equipment and storage medium | |
| Chang et al. | CT manifestations of gallbladder carcinoma based on neural network | |
| Alam et al. | Teeth segmentation by optical radiographic images using VGG-16 deep learning convolution architecture with R-CNN network approach for biomedical sensing applications | |
| Santhi et al. | Oral disease detection from dental X-Ray images using Densenet | |
| CN112420171A (en) | Maxillary sinus floor bone classification method and system based on artificial intelligence | |
| Hasnain et al. | Deep learning architectures in dental diagnostics: a systematic comparison of techniques for accurate prediction of dental disease through x-ray imaging | |
| CN115481681B (en) | Mammary gland sampling data processing method based on artificial intelligence | |
| Tanwir et al. | Artificial intelligence in dentistry: a glance to the future | |
| Velusamy et al. | Faster Region‐based Convolutional Neural Networks with You Only Look Once multi‐stage caries lesion from oral panoramic X‐ray images | |
| Bhatnagar | Identification of Dental Implants with the Development of an Artificial Intelligence Model using Radiograph | |
| Singh et al. | To Use a Quick R-CNN Algorithm—An Automated Strategy for Tooth Diagnostics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |